From a3d2b4e357257d99c100225f4c10e4bf291082e9 Mon Sep 17 00:00:00 2001
From: Even Rouault <even.rouault@spatialys.com>
Date: Sat, 20 Jul 2024 02:20:49 +0200
Subject: [PATCH] XLSX: support documents whose XML elements have a prefix

Fixes https://github.com/duckdb/duckdb_spatial/issues/362
---
 autotest/ogr/data/xlsx/with_xml_prefix.xlsx | Bin 0 -> 2457 bytes
 autotest/ogr/ogr_xlsx.py                    |  15 +++++++
 ogr/ogrsf_frmts/xlsx/ogrxlsxdatasource.cpp  |  45 ++++++++++++++------
 3 files changed, 46 insertions(+), 14 deletions(-)
 create mode 100644 autotest/ogr/data/xlsx/with_xml_prefix.xlsx

diff --git a/autotest/ogr/data/xlsx/with_xml_prefix.xlsx b/autotest/ogr/data/xlsx/with_xml_prefix.xlsx
new file mode 100644
index 0000000000000000000000000000000000000000..4d23ef8ddf26261784663b877b684c626a460d78
GIT binary patch
literal 2457
zcmWIWW@Zs#U|`^2C|~h9;;~b|)q5cCE+YejFoO(3v~zx5Norn6d`M+MYH_SyMQ%=L
z2qy#cB%KLqi+{~ZE3M#WU}Sm0%)kI9)=sqbJ7OTvx;|;A>&FyL0VmIG5gCn5+!5z2
zHeRd$WZ`Wg&FcP$t75MP>mtWRQa3-}`<?dY{P+I6{Z^;_W}INT|9NhPrStkj8Go*P
zR(<_!+WYo*Pj!_xDt*23+46Y>yF-?nx@SwrzLF9Tzh^~0!Y(WOikx=j-IfkGnEqww
z$>%(FCEr&(O){>IIQ^A1`pq0!y<Ntuo*TJkPdT%cSKa<!gNbAI;sTzm!lOzWYEzWY
zcHE2%TwQSK$7(5_{YJu3zofrZYX{B=b>7^$MRDFKWi?}^>S>zxUg4h-wtkShw2{vx
z;I!YL_nV^snQ>e>FSU;$02p>$z|dk~*u?uK0?dFCY{0OKFG|fR)&~dI0$@;8GtNd2
zE|7ji*mtOXjNb{=uK^5uZn%CukSXvm|A%Qv?|I+6Lk0q@4^~HPR!^R^ic5n}+AK#<
z+~Ax==cEdq+kd{~#D4ME%M{1-YSHQCm7hJ|fB(Jw=H2O$F`7pv&TT4Hbev#(r*~a=
zfoECh;dQxvCa<SR7fRHM{5<N)>g-k5EpgVNyN|WH=x36pj@46%sxP;EcB$(u{d43)
z&VkjMe)04EDR0`kc%kj=(@uJS8TXqy%qe-Hsd?ack2&AvnUz|bxE@{5v<SO&?e)4l
zA)D6NnU!)f-DZpbwv8VVdcs1ep~npLeMJr==$-+6jK$Z8paW?ROX6eS3Jf|4V9@bH
zG?(WWWhdq5XTy^$4ug98ZTSuv@U)iu305xG<lzYY8yR8YmT}iCOW3`TYp*mjFuk@k
zy*Tycy#1%@_m%ZeuI9KEx}Kkv!?;_8wQ<`rx7+I)G`HMV4oS63H~)0FHACRzy&^O5
z3+)e#G<UwcsNr{FVejr~djb`=?VJ3@W>s7Dkp#D${+F9gyrL`qM=$%MoH+F$lNj5^
zDaFEn8$QhJTOj>I>v#Q>mAR$8>Z{b7&1#m|_GulNawq$^Ldwk3SKDHQr_B~e1RW<@
z(D5OBU!0MeS^`agmk5O-JJ6hntiUDTfuZ=72^f+X=75bega_wY>?WQKI-Pe|L152s
zZGrDBx7$~3N?K(0x54W#)1wCxM>mM&<kr_QdUra{maBTzJ9BO=&+grS*HjyMy%w1<
zOYHU>p-d?Y&Ej=0)wj(~-}U@v-a54n;@3SpkEBb4pEZ3e(B5Tuy(EZxd#j_AeD=Yt
zM!8$Y#m!S6iCiu@f6AslDN4h+(ye2?By&EK>w$?I4Fq35U%)NAQ_Vo1RW%|wwJP-q
z`>oXqUQ>1-TzXDQ|4j8E*Ka3;6n{IuEIq-0L0qw`eD00d)eQNIS*|Rxi~HZ$|LMHl
zBEDrMu}{C3PQAFyo40QdD@W+D`}Hpqe+Axff4TbMq2kgQ#<=&>)8i2lVucnVoREM-
zDh%|QW~UheF<RM)NYNl8UP#M6*a(aZIbe#GLN)?9T_ef{keMKiW@hhsLr~&oy)ZY}
z<=)C@ww8d)dOB-Xh<5}n@6DL3;kj^5Z=-S{tCHA?BL`A$-mjZ9|GRtS`Ns$Jo-}W3
zbDsG0zzL7rS(giEEQ)%+VBINUUZrKd7qZ18tPC5S<rm*pX1jS=!zoR6<`j+B?t;Ou
ztxKHdxw`K-`C0c@KJzrS?7Y5zf@_1KE*(6TsUlo|pknJknW{w`YXtW#KEe3>p)ubN
zJ=bX|3CE{3=QSpVK6-RA>nl@$HzSiAGonmHt&b#t0m8rl3=xJcjUX1*+KClXJE4UH
z#9&;7HN;>B1_6e@jy6E^VX*}?3siJtH4EWL)Pft?ku`9$pixEGahPQxvg4|NSrc3`
z0u2HcXn0(PSuP^GY&*mt<kS=zg3nM$&|wxg$gbN8YzF{c2Qn01_yElW6+U<zidpm^
zJM=GxnaBkZ&~#8igw=G2i!t*fvWwkV!7fIMZ!Eb8IZ<NfO=QP*A{&PswutD*YCbC)
OC^>TiA*k%_1M>iMUMDaB

literal 0
HcmV?d00001

diff --git a/autotest/ogr/ogr_xlsx.py b/autotest/ogr/ogr_xlsx.py
index 197563a1c964..632df94bcfab 100755
--- a/autotest/ogr/ogr_xlsx.py
+++ b/autotest/ogr/ogr_xlsx.py
@@ -650,3 +650,18 @@ def test_ogr_xlsx_write_sheet_without_row():
     assert ds.GetLayer(2).GetFeatureCount() == 1
     ds = None
     gdal.Unlink(tmpfilename)
+
+
+###############################################################################
+# Test reading a XLSX file with XML element prefixes
+
+
+def test_ogr_xlsx_read_xml_prefix():
+
+    ds = ogr.Open("data/xlsx/with_xml_prefix.xlsx")
+    lyr = ds.GetLayer(0)
+    assert lyr.GetLayerDefn().GetFieldDefn(0).GetName() == "Col1"
+    assert lyr.GetLayerDefn().GetFieldDefn(1).GetName() == "Col2"
+    f = lyr.GetNextFeature()
+    assert f["Col1"] == "foo"
+    assert f["Col2"] == "bar"
diff --git a/ogr/ogrsf_frmts/xlsx/ogrxlsxdatasource.cpp b/ogr/ogrsf_frmts/xlsx/ogrxlsxdatasource.cpp
index 7222ca2ef468..9590b7b5f20b 100644
--- a/ogr/ogrsf_frmts/xlsx/ogrxlsxdatasource.cpp
+++ b/ogr/ogrsf_frmts/xlsx/ogrxlsxdatasource.cpp
@@ -418,6 +418,18 @@ int OGRXLSXDataSource::Create(const char *pszFilename,
     return TRUE;
 }
 
+/************************************************************************/
+/*                           GetUnprefixed()                            */
+/************************************************************************/
+
+static const char *GetUnprefixed(const char *pszStr)
+{
+    const char *pszColumn = strchr(pszStr, ':');
+    if (pszColumn)
+        return pszColumn + 1;
+    return pszStr;
+}
+
 /************************************************************************/
 /*                           startElementCbk()                          */
 /************************************************************************/
@@ -434,6 +446,8 @@ void OGRXLSXDataSource::startElementCbk(const char *pszNameIn,
     if (bStopParsing)
         return;
 
+    pszNameIn = GetUnprefixed(pszNameIn);
+
     nWithoutEventCounter = 0;
     switch (stateStack[nStackDepth].eVal)
     {
@@ -474,6 +488,8 @@ void OGRXLSXDataSource::endElementCbk(const char *pszNameIn)
     if (bStopParsing)
         return;
 
+    pszNameIn = GetUnprefixed(pszNameIn);
+
     nWithoutEventCounter = 0;
 
     nDepth--;
@@ -1346,6 +1362,8 @@ void OGRXLSXDataSource::startElementSSCbk(const char *pszNameIn,
     if (bStopParsing)
         return;
 
+    pszNameIn = GetUnprefixed(pszNameIn);
+
     nWithoutEventCounter = 0;
     switch (stateStack[nStackDepth].eVal)
     {
@@ -1381,11 +1399,14 @@ static void XMLCALL endElementSSCbk(void *pUserData, const char *pszNameIn)
     ((OGRXLSXDataSource *)pUserData)->endElementSSCbk(pszNameIn);
 }
 
-void OGRXLSXDataSource::endElementSSCbk(CPL_UNUSED const char *pszNameIn)
+void OGRXLSXDataSource::endElementSSCbk(const char * /*pszNameIn*/)
 {
     if (bStopParsing)
         return;
 
+    // If we were to use pszNameIn, then we need:
+    // pszNameIn = GetUnprefixed(pszNameIn);
+
     nWithoutEventCounter = 0;
 
     nDepth--;
@@ -1529,6 +1550,8 @@ void OGRXLSXDataSource::startElementWBRelsCbk(const char *pszNameIn,
     if (bStopParsing)
         return;
 
+    pszNameIn = GetUnprefixed(pszNameIn);
+
     nWithoutEventCounter = 0;
     if (strcmp(pszNameIn, "Relationship") == 0)
     {
@@ -1593,18 +1616,6 @@ void OGRXLSXDataSource::AnalyseWorkbookRels(VSILFILE *fpWorkbookRels)
     VSIFCloseL(fpWorkbookRels);
 }
 
-/************************************************************************/
-/*                           GetUnprefixed()                            */
-/************************************************************************/
-
-static const char *GetUnprefixed(const char *pszStr)
-{
-    const char *pszColumn = strchr(pszStr, ':');
-    if (pszColumn)
-        return pszColumn + 1;
-    return pszStr;
-}
-
 /************************************************************************/
 /*                          startElementWBCbk()                         */
 /************************************************************************/
@@ -1621,8 +1632,10 @@ void OGRXLSXDataSource::startElementWBCbk(const char *pszNameIn,
     if (bStopParsing)
         return;
 
+    pszNameIn = GetUnprefixed(pszNameIn);
+
     nWithoutEventCounter = 0;
-    if (strcmp(GetUnprefixed(pszNameIn), "sheet") == 0)
+    if (strcmp(pszNameIn, "sheet") == 0)
     {
         const char *pszSheetName = GetAttributeValue(ppszAttr, "name", nullptr);
         const char *pszId = GetAttributeValue(ppszAttr, "r:id", nullptr);
@@ -1725,6 +1738,8 @@ void OGRXLSXDataSource::startElementStylesCbk(const char *pszNameIn,
     if (bStopParsing)
         return;
 
+    pszNameIn = GetUnprefixed(pszNameIn);
+
     nWithoutEventCounter = 0;
     if (strcmp(pszNameIn, "numFmt") == 0)
     {
@@ -1810,6 +1825,8 @@ void OGRXLSXDataSource::endElementStylesCbk(const char *pszNameIn)
     if (bStopParsing)
         return;
 
+    pszNameIn = GetUnprefixed(pszNameIn);
+
     nWithoutEventCounter = 0;
     if (strcmp(pszNameIn, "cellXfs") == 0)
     {